There are many discussions here on how humans can work along AI, but what is stopping Mathematics from becoming an exclusively AI practice?
- It doesn't seem the capabilities of AI will plateau any time soon.
- I have heard the use of AI will allow Mathematicians to work on big picture stuff, but we cannot even solve the conjectures AI is currently settling.
- It can check its proofs with automated provers.
- AIs are much better at filtering through large amounts of information, so there is no reason not to "publish" every result they find.
- If Math becomes purely AI driven, there will be no need for mathematicians to teach Math, and no one for them to teach Math to.
Couldn't we be a couple of years (months?) away from AI deciding which problems it wants to solve, solving them, getting its proofs checked in Lean, and publishing them in an online database for other AI to check? All in maybe hours, while our current process takes months to years.
source & further reading
mathoverflow.net — original article
Why are you not panicking? [closed]
Examples of problems that current AI models fail to solve
Since AI (Claude) just helped find the new infinite family $a^4+b^4+c^4+d^4 = (a+27b+27c+27d)^4$, should we expect more similar discoveries soon?